SoftBank Corp has increased its holdings in Nvidia and TSMC, regulatory filings show.

The Japanese conglomerate raised its stake in Nvidia to about US$3 billion by the end of March 2025, up from US$1 billion in the previous quarter, and acquired roughly US$330 million in TSMC shares and US$170 million in Oracle shares.

SoftBank’s Vision Fund also sold nearly US$2 billion in public and private assets in the first half of 2025.

SoftBank, led by founder Masayoshi Son, is focusing on AI-related hardware and has been building a portfolio around UK-based chip designer Arm Holdings.

The company sold a 4.9% stake in Nvidia in 2019, which would be worth over US$200 billion, but has since sought to regain exposure to the semiconductor sector.

SoftBank has also raised about US$4.8 billion by selling part of its T-Mobile stake in June 2025, which may support further hardware investments.

🔗 Source: Bloomberg

🧠 Food for thought

1️⃣ SoftBank’s Nvidia misstep highlights the massive cost of poor timing in tech investments

SoftBank’s current $3 billion Nvidia stake pales in comparison to what they gave up just six years ago.

The company sold a 4.9% stake in Nvidia in early 2019 that would be worth more than $200 billion today, creating one of the most expensive investment mistakes in tech history1.

This timing mishap occurred because Vision Fund losses hampered SoftBank’s ability to be an early investor in generative AI, forcing them to exit positions before the AI boom took off.

The scale of this missed opportunity becomes clearer when considering that SoftBank’s entire market capitalization today stands at around $119 billion, less than what their old Nvidia stake alone would be worth.

Now Son is attempting to rebuild AI exposure through multiple smaller bets, including stakes in TSMC ($330 million) and Oracle ($170 million), but these investments represent a fraction of the wealth creation they missed.

2️⃣ SoftBank’s AI infrastructure push aligns with growing geopolitical semiconductor tensions

SoftBank’s timing on AI infrastructure investments coincides with intensifying US-China competition over semiconductor supremacy.